微信小程序页面迁移校验之前 P5任务处理之前

apps/backend/app/services/task_executor.py

@@ -17,19 +17,24 @@ from __future__ import annotations
 import asyncio
 import logging
 import subprocess
-import sys
 import threading
 import time
 from datetime import datetime, timezone
 from typing import Any
 
-from ..config import ETL_PROJECT_PATH
+# CHANGE 2026-03-07 | 只保留模块引用，execute() 中实时读取属性值
+# 禁止 from ..config import ETL_PROJECT_PATH（值拷贝，reload 后过期）
+from .. import config as _config_module
 from ..database import get_connection
 from ..schemas.tasks import TaskConfigSchema
 from ..services.cli_builder import cli_builder
 
 logger = logging.getLogger(__name__)
 
+# 实例标识：用于区分多后端实例写入同一 DB 的记录
+import platform as _platform
+_INSTANCE_HOST = _platform.node()  # hostname
+
 
 class TaskExecutor:
     """管理 ETL CLI 子进程的生命周期"""
@@ -112,21 +117,58 @@ class TaskExecutor:
         execution_id: str,
         queue_id: str | None = None,
         site_id: int | None = None,
+        schedule_id: str | None = None,
     ) -> None:
         """以子进程方式调用 ETL CLI。
 
         使用 subprocess.Popen + 线程读取，兼容 Windows（避免
         asyncio.create_subprocess_exec 在 Windows 上的 NotImplementedError）。
         """
+        # CHANGE 2026-03-07 | 实时从 config 模块读取，避免 import 时复制的值过期
+        etl_path = _config_module.ETL_PROJECT_PATH
+        etl_python = _config_module.ETL_PYTHON_EXECUTABLE
+
         cmd = cli_builder.build_command(
-            config, ETL_PROJECT_PATH, python_executable=sys.executable
+            config, etl_path, python_executable=etl_python
         )
         command_str = " ".join(cmd)
+
+        # CHANGE 2026-03-07 | 运行时防护：拒绝执行包含非预期路径的命令
+        # 检测两种异常：
+        # 1. D 盘路径（junction 穿透）
+        # 2. 多环境子目录（test/repo、prod/repo）
+        _cmd_normalized = command_str.replace("/", "\\")
+        _bad_patterns = []
+        if "D:\\" in command_str or "D:/" in command_str:
+            _bad_patterns.append("D盘路径")
+        if "\\test\\repo" in _cmd_normalized or "\\prod\\repo" in _cmd_normalized:
+            _bad_patterns.append("多环境子目录(test/repo或prod/repo)")
+
+        if _bad_patterns:
+            _issues = " + ".join(_bad_patterns)
+            logger.error(
+                "路径防护触发：命令包含 %s，拒绝执行。"
+                " command=%s | ETL_PY=%s | ETL_PATH=%s"
+                " | NEOZQYY_ROOT=%s | config.__file__=%s",
+                _issues, command_str, etl_python, etl_path,
+                __import__('os').environ.get("NEOZQYY_ROOT", "<未设置>"),
+                _config_module.__file__,
+            )
+            raise RuntimeError(
+                f"ETL 命令包含异常路径（{_issues}），拒绝执行。"
+                f" 请检查 .env 中 ETL_PYTHON_EXECUTABLE 和 ETL_PROJECT_PATH 配置。"
+                f" 当前值: ETL_PY={etl_python}, ETL_PATH={etl_path}"
+            )
+
         effective_site_id = site_id or config.store_id
 
+        # CHANGE 2026-03-07 | 在 command 前缀中注入实例标识，
+        # 便于在多后端实例共享同一 DB 时区分记录来源
+        command_str_with_host = f"[{_INSTANCE_HOST}] {command_str}"
+
         logger.info(
             "启动 ETL 子进程 [%s]: %s (cwd=%s)",
-            execution_id, command_str, ETL_PROJECT_PATH,
+            execution_id, command_str, etl_path,
         )
 
         self._log_buffers[execution_id] = []
@@ -140,7 +182,8 @@ class TaskExecutor:
             task_codes=config.tasks,
             status="running",
             started_at=started_at,
-            command=command_str,
+            command=command_str_with_host,
+            schedule_id=schedule_id,
         )
 
         exit_code: int | None = None
@@ -226,7 +269,7 @@ class TaskExecutor:
             cmd,
             stdout=subprocess.PIPE,
             stderr=subprocess.PIPE,
-            cwd=ETL_PROJECT_PATH,
+            cwd=_config_module.ETL_PROJECT_PATH,
             env=env,
             text=True,
             encoding="utf-8",
@@ -302,18 +345,30 @@ class TaskExecutor:
         status: str,
         started_at: datetime,
         command: str,
+        schedule_id: str | None = None,
     ) -> None:
         """插入一条执行日志记录（running 状态）。"""
         try:
             conn = get_connection()
             try:
                 with conn.cursor() as cur:
+                    # 如果调用方未传 schedule_id，尝试从 task_queue 回查
+                    effective_schedule_id = schedule_id
+                    if effective_schedule_id is None and queue_id is not None:
+                        cur.execute(
+                            "SELECT schedule_id FROM task_queue WHERE id = %s",
+                            (queue_id,),
+                        )
+                        row = cur.fetchone()
+                        if row and row[0]:
+                            effective_schedule_id = str(row[0])
+
                     cur.execute(
                         """
                         INSERT INTO task_execution_log
                             (id, queue_id, site_id, task_codes, status,
-                             started_at, command)
-                        VALUES (%s, %s, %s, %s, %s, %s, %s)
+                             started_at, command, schedule_id)
+                        VALUES (%s, %s, %s, %s, %s, %s, %s, %s)
                         """,
                         (
                             execution_id,
@@ -323,6 +378,7 @@ class TaskExecutor:
                             status,
                             started_at,
                             command,
+                            effective_schedule_id,
                         ),
                     )
                 conn.commit()